Erythropoietin-Induced Activation of the JAK2/STAT5, PI3K/ Akt, and Ras/ERK Pathways Promotes Malignant Cell Behavior in a Modified Breast Cancer Cell Line

Published OnlineFirst March 30, 2010; DOI: 10.1158/1541-7786.MCR-09-0264

Molecular Signaling and Regulation Cancer Research Erythropoietin-Induced Activation of the JAK2/STAT5, PI3K/ Akt, and Ras/ERK Pathways Promotes Malignant Cell Behavior in a Modified Breast Cancer Cell Line

Zhanzhong Shi1, Vivien M. Hodges1, Elaine A. Dunlop4, Melanie J. Percy3, Alexander P. Maxwell2, Mohamed El-Tanani1, and Terry R.J. Lappin1

Abstract Erythropoietin (Epo), the major regulator of erythropoiesis, and its cognate receptor (EpoR) are also expressed in nonerythroid tissues, including tumors. Clinical studies have highlighted the potential adverse effects of erythropoiesis-stimulating agents when used to treat cancer-related anemia. We assessed the ability of EpoR to enhance tumor growth and invasiveness following Epo stimulation. A benign noninvasive rat mammary cell line, Rama 37, was used as a model system. Cell signaling and malignant cell behavior were compared between parental Rama 37 cells, which express few or no endogenous EpoRs, and a modified cell line stably transfected with human EpoR (Rama 37-28). The incubation of Rama 37-28 cells with pharmacologic levels of Epo led to the rapid and sustained increases in phosphorylation of signal transducers and activators of transcription 5, Akt, and extracellular signal-regulated kinase. The activation of these signaling pathways significantly increased invasion, migration, adhesion, and colony formation. The Epo-induced invasion capacity of Rama 37-28 cells was reduced by the small interfering RNA–mediated knockdown of EpoR mRNA levels and by inhibitors of the phosphoi- nositide 3-kinase/Akt and Ras/extracellular signal-regulated kinase signaling pathways with adhesion also reduced by Janus-activated kinase 2/signal transducers and activators of transcription 5 inhibition. These data show that Epo induces phenotypic changes in the behavior of breast cancer cell lines and establishes links between individ- ual cell signaling pathways and the potential for cancer spread. Mol Cancer Res; 8(4); 615–26. ©2010 AACR.

Introduction to determine whether EpoR activation modifies tumor cell growth. Erythropoietin (Epo) is a glycoprotein hormone, pro- Recombinant human Epo treatment has transformed the duced by the kidney in response to hypoxia, and acts quality of life for millions of anemia sufferers, especially through its cognate receptor (EpoR) to trigger signaling those with chronic kidney disease. More recently, erythro- cascades that result in proliferation, differentiation, and poiesis-stimulating agents (ESA) have been used to treat survival of erythroid progenitors (reviewed in refs. 1, 2). the cancer-related anemia, but there is emerging evidence Following the detection of expression of Epo and its recep- that such therapies may be harmful to some groups of can- tor on multiple cell types, it has become clear that Epo has cer patients. In 2003, Henke and colleagues (17) described pleiotropic effects extending well beyond the maintenance the outcome of Epo treatment in a randomized trial of 351 of red cell mass (3). In recent years, multiple investigators anaemic head and neck cancer patients undergoing radio- have documented the presence of EpoR expression in nu- therapy. Unexpectedly, locoregional progression-free survival was worse in the Epo treatment arm compared with merous tumor cell lines and carcinomata (4-14) but the placebo. In the same year, the Breast Cancer Erythropoie- specificity of commercially available EpoR antibodies has tin Survival Trial reported that overall survival was poorer been questioned (15, 16). Functional studies are needed in the patients assigned to the Epo treatment arm (18). A Cochrane Review collated data on 9,353 cancer patients in 57 trials in which ESAs were given to prevent or treat Authors' Affiliations: 1Centre for Cancer Research and Cell Biology and 2Genetic Epidemiology Research Group, Queen's University Belfast, and anemia (19). There was a significantly higher relative risk 3Department of Haematology, Belfast City Hospital, Belfast, Northern for thromboembolic events in ESA-treated patients com- Ireland, United Kingdom; and 4Institute of Medical Genetics, Cardiff pared with controls (relative risk, 1.67; confidence interval, University, Cardiff, South Glamorgan, United Kingdom 1.35-2.06; ref. 20). A recent Food and Drug Administra- Note: Z. Shi and V.M. Hodges contributed equally to this work. tion alert has highlighted concerns about the potential se- Corresponding Author: Terry R. Lappin, Queen's University Belfast, 97 rious adverse effects of ESA treatment of anemic cancer Lisburn Road, Belfast, Northern Ireland BT9 7BL, United Kingdom. patients and postulated that this could be due to an in- Phone: 028-90972929; Fax: 028-90972776. E-mail: [email protected] creased risk of thrombosis, tumor growth, and/or neovas- doi: 10.1158/1541-7786.MCR-09-0264 cularisation (21, 22). These regulatory concerns have led to ©2010 American Association for Cancer Research. further revision of the clinical guidelines for use of ESAs in

www.aacrjournals.org 615

Shi et al.

anaemic cancer patients (23). Recently, two meta-analysis AGA GCA AGC CAC ATA (Eurofins MWG Operon). reports that included data from a total of 13,933 cancer For the production of stable transformant cell lines, Rama patients enrolled in 53 trials using ESAs (24) and 52 trials 37 cells were seeded in six-well plates at 1 × 105/mL in with 12,009 patients (25) have both concluded that ESAs antibiotic-free medium. Rama 37 cells were transfected are associated with an increased risk of death with active with an expression vector pcDNA3.1 for wild-type human treatment and worse overall survival. EpoR and selected in 1.0 mg/mL geneticin (Invitrogen) as – EpoR, a 508 amino acid transmembrane protein, is a type previously described (31). Single-cell clones of transfor- I cytokine receptor belonging to a family of proteins that mants, which overexpressed the transfected EpoR as deter- includes the prolactin, growth hormone, and interleukin-3 mined by quantitative real-time reverse transcriptase-PCR receptors. After Epo binding, preformed Epo receptor dimers and Western blot screening, were generated. Two clones, undergo a conformational change that activates three major Rama 37-28 and Rama 37-50, with significantly increased – signal transduction pathways: the Janus-activated kinase EpoR mRNA levels, were selected for further study. signal transducer and activator of transcription (JAK2- Transient transfection of human EpoR small interfering RNA STAT5), phosphatidylinositol-3-kinase-Akt (PI3K-Akt), (siRNA; Ambion) into Rama 37-28 and MDA-MB-435s and Ras-extracellular signal-related kinase (Ras-ERK) cas- cell lines was carried out in six-well plates using the siPORT cades (reviewed in ref. 1). NeoFX transfection agent according to the manufacturer's – Recent studies using the non small cell lung carcinoma instructions (Ambion). Briefly, before the transfection, cells cell line H838 have confirmed that Epo at pharmacologic were trypsinized and suspended in media without antibio- levels can activate the PI3K/Akt, JAK2/STAT5, and Ras/ tics at a cell density of 1 × 105/mL. SiPORT NeoFX (5 μL) ERK signaling cascades in nonerythroid cells (26). Down- was diluted into Opti-MEM medium (95 μL; Invitrogen), regulation of Epo-induced signaling was found to be im- incubated at room temperature for 10 min, then mixed paired in these cells (27), suggesting that Epo may with an equal volume of appropriately diluted EpoR siRNA provide a growth advantage to tumor cells. To further in- solution (final concentration, 10 nmol/L). After incubation vestigate the ability of Epo to enhance tumor growth and at room temperature for 10 min, siRNA transfection com- the potential for metastatic spread, we have compared Ra- plexes (200 μL) were dispensed into a six-well plate. Cell ma 37, a benign noninvasive rat mammary cell line expres- suspensions (2.3 mL) were overlaid onto the transfection sing low levels of endogenous EpoR, to a modified Rama complexes, gently mixed, and incubated for 72 h at 37°C, 37 cell line stably transfected with human EpoR to express 5% CO2. a higher level of receptor (Rama 37-28). Both cell types, Rama 37 and Rama 37-28 expressing low and high levels Quantitative Real-time Reverse Transcriptase-PCR of EpoR, respectively, were incubated with pharmacologic EpoR mRNA expression was measured in the parental levels of Epo (10 U/mL). Here, we present evidence that cells and transfected cell lines by Q-PCR. Briefly, total Epo signaling through the JAK2/STAT5, PI3K/Akt, and RNA was extracted from cell lines with Trizol reagent Ras/ERK pathways can mediate an increase in malignant (Invitrogen). Following reverse transcription, cDNA was cell behavior in the presence of EpoR. amplified by Taqman probe–based chemistry as previously described (32) using predesigned primer/probe sets for Materials and Methods human and rat EpoR (human: Hs 00181092.m1; rat: Rn00566533_m1; Applied Biosystems). For comparison, Cell Lines and Cell Culture a primer/probe set for the endogenous control 18S RNA, The rat mammary Rama 37 nonmetastatic benign tumor- which recognizes human, mouse, and rat transcripts, was derived cell line (28), the derivative stably transformed used (Applied Biosystems). Samples that gave an 18S CT cell subclones, and MDA-MB-435s human breast epi- value outside the range of 10 ± 3 were excluded from Δ thelial cell line (29, 30) were cultured in DMEM with further analysis. Relative Q-PCR EpoR CT values were high glucose (4.5 g/L), 10% (v/v) FCS, 100 μg/mL calculated by comparison with 18S levels. A standard penicillin, and 100 μg/mL streptomycin (Invitrogen) at curve using linearized EpoR in pCDNA3.1 plasmid 37°C in an atmosphere of 5% (v/v) CO2.Erythroid DNA was generated to calculate the number of EpoR progenitors were derived from PBSC harvests, frozen transcripts per 10-ng total RNA from a real-time PCR pre-2005, and consented for research at that time. Cells CT value. The equation generated from the standard 5 − (6 × 10 per mL) were plated in triplicate in HSC- curve was CT value = 3.012 log (EpoR copy number) CFUlite methylcellulose medium with EPO (Miltenyi + 34.65. Biotec). Blast-forming unit (erythroid) colonies were harvested after 12 d of culture at 37°C. Western Blot Analysis Western blot analysis was done as previously described Stable and Transient Transfections (26) with minor modifications. For the detection of EpoR Human EpoR was amplified from cDNA and cloned in- expression, cells were harvested by trypsinisation and were to the pcDNA3.1/V5-His-TOPO using the following pri- lysed in radioimmunoprecipitation assay buffer. Cells for mers: forward, TTT TTT AAG CTT ATG GAC CAC signaling pathway immunoblotting analysis were plated CTC GGG GCG; reverse, TTT TTT TTG AAT TCC atadensityof1×105/mL and incubated overnight.

616 Mol Cancer Res; 8(4) April 2010 Molecular Cancer Research

Erythropoietin-Induced Malignant Cell Behavior

Following three washes with PBS, the cells were serum coated with Matrigel. The percentage of migration of each starved for 24 h and then, on the day of the experiment, cell type was normalized using the percentage of migration were treated with Epo (10, 2, or 0.2 U/mL) for 0, 5, 15, of parental Rama 37 cells without Epo treatment as 100%, 30, 60, and 120 min with additional controls without Epo as outlined above. for 30- and 60-min incubation times. Whole-cell lysates were obtained by harvesting cells into a Laemmli buffer Adhesion Assay followed by sonication and boiling for 5 min. Samples Cell adhesion assays were done as previously described were separated on 10% polyacrylamide gels; transferred (31) with minor modifications. Briefly, cells were plated to polyvinylidene difluoride membranes; and probed with at 2 × 105 cells per well in a six-well plate coated with Ma- antibodies for pSTAT5 (Upstate), STAT5, pAkt, Akt, trigel (BD Biosciences) in conditioned media, treated with pERK 1/2, and ERK 1/2 (all from Cell Signaling, New or without 10 U/mL Epo, and incubated at 37°C for 18 h in England Biolabs). Detection was done by horseradish per- an atmosphere of 5% (v/v) CO2. Cells were washed with oxidase–conjugated secondary antibodies (DAKO) and en- PBS, fixed with methanol, stained with 0.5% crystal violet, hanced chemiluminescence plus Western Blotting washed with distilled water, and air dried. Sodium citrate/ Detection System (Amersham Biosciences). For the signal- ethanol solution was added and the cells were gently agitated ing pathway inhibition experiments, cells were pretreated for 30 min. An aliquot (50 μL) from each well was trans- with JAK inhibitor 1, U0126, or PI-103 (all from Calbio- ferred into a 96-well plate and read at 570 nm in a Tecan chem, Merck), at optimized final concentrations of 5, 20, plate reader. The percentage of adhesion of each cell type and 0.5 μmol/L, respectively, for 1 h before the addition of was normalized using the percentage of adhesion of parental Epo. Cell lysates prepared by direct addition of Laemmli Rama 37 cells without Epo treatment as 100%. buffer to cells at 0-, 15-, 30-, and 60-min time points were retained at −80°C. Colony Formation Assay Colony formation assays were done on soft agar. Base In vitro Tests for Cell Adhesion, Invasion, and agar was prepared by mixing an equal volume of 1% agar Colony Formation with 2 × DMEM culture media containing high glucose Assays for cell adhesion, colony formation, and invasion (4.5 g/L), 10% (v/v) FCS, 100 μg/mL penicillin, and through Matrigel in Boyden chambers were carried out as 100 μg/mL streptomycin (Invitrogen). A 1.5-mL aliquot previously described (33). Cells for adhesion, migration, of this mixture was added to a 35-mm Petri dish and invasion, and colony formation assays were washed thrice allowed to set. Cells (2 × 104) were seeded into 10-mL cen- with PBS and then plated in serum-free media for 24 h be- trifuge tubes containing 3 mL of 2× DMEM culture media fore treatment with Epo. and 3 mL of 0.7% Agar. The solution was mixed gently and 1.5 mL were added on top of the base agar in tripli- Invasion Assay cate. Plates were incubated at 37°C in a humidified incu- Invasion was measured using Matrigel-coated multiwell bator with 5% CO2 for 10 to 14 d. Plates were stained inserts as previously described (31) with minor modifica- with 0.5 mL of 0.005% crystal violet for over 1 h and col- tions. Matrigel-coated invasion chambers (6.4-mm diame- onies were counted using a dissecting microscope. ter; 8-μm pore size; BD Biosciences) were used to assess the invasive capacity of parental Rama 37 and EpoR-trans- Results fected Rama 37-28 and Rama 37-50 cell lines. Briefly, 1 × 105 cells were resuspended in serum-free, phenol red–free Epo Receptor Overexpression DMEM media and placed in the upper invasion chambers. EpoR mRNA expression in Rama 37 and Rama 37-28 The cultures were incubated with or without Epo treat- cells was quantified by Q-PCR using both human- and ment with or without kinase inhibitors for 18 h at 37°C rat-specific primer probe sets (Table 1) and immunoblot- in a 5% (v/v) CO2 atmosphere. The upper surfaces of ting (Fig. 1A). In the parental Rama 37 cells, EpoR levels the filters were wiped clean of cells and the filters were then were barely detectable (CT > 35). By comparison, high le- fixed by immersion in 100% (v/v) methanol and stained vels of expression for both EpoR mRNA and protein were by 0.5% crystal violet for 25 min. Each membrane was observed in Rama 37-28 cells with stable transfection of washed in running distilled water and left to air dry. After EpoR. This model system permitted direct comparison of washing with sodium citrate/ethanol for 30 min, 200 μL benign noninvasive mammary epithelial cells expressing of the solution were transferred into a 96-well plate and very low EpoR with cells expressing an increased number read at 570 nm in a Tecan plate reader. The percentage of receptors. To investigate the effect of Epo on the cellular of invasion of each cell type was normalized using the per- properties of parental Rama 37 cells compared with EpoR- centage of invasion of parental Rama 37 cells without Epo transfected Rama 37-28 cells, a series of invasion, adhesion, treatment as 100%. migration, and colony formation assays were done in the presence or absence of Epo. Epo treatment of parental Migration Assay Rama 37 cells did not affect invasion, adhesion, or migra- The migration assay was similar to the invasion assay de- tion. However, Rama 37-28 cells stably transfected with scribed above except that the insert membrane was not EpoR showed significantly increased invasion (P < 0.001),

www.aacrjournals.org Mol Cancer Res; 8(4) April 2010 617

Shi et al.

Activation of Three Signaling Pathways by Table 1. EpoR mRNA expression level as Erythropoietin determined by Q-PCR and expressed as To assess EpoR signaling activity, Epo activation of the corrected CT three pathways was investigated. Phosphorylation of STAT5, Akt, and ERK1/2 were monitored over a period Cell line Human EpoR Rat EpoR of 2 hours after Epo stimulation by immunoblotting. Pa- CT value CT value rental Rama 37 cells (serum starved for 24 hours) showed no activation of the JAK/STAT pathway and only minor, Rama 37 >35 >35 short-lived activation of the PI3K/Akt and Ras/ERK Rama 37-28 19.28 ± 0.43 >35 pathways (Fig. 3A). In contrast, Epo stimulation of Rama 37-28 cells resulted in the significant and sustained activation of all three signaling pathways (Fig. 3B). Activation P P adhesion ( < 0.001), and migration ( < 0.01) in response to was observed only 5 minutes following stimulation with Epo stimulation (Fig. 1B-D) compared with nontreated con- maximal levels seen at 15 minutes (highlighted). The level trols. The number of Rama 37-28 colonies also increased sig- of total STAT5, Akt, and ERK1/2 protein remained P nificantly in the presence of Epo ( < 0.001), but there was no unchanged. This confirmed that the transfected EpoR in change in the colony-forming potential in the parental Rama Rama 37-28 cells was functionally active in response to 37 cells exposed to Epo (Fig. 1E). Invasion capacity was also Epo and activation of these pathways mediated a signifi- tested with a second stably transfected clone with a lower level cant shift toward a metastatic cell phenotype. Signaling of Epo receptors (Rama 37-50; EpoR CT, 22.8 ± 0.14) and pathways are still activated in these cells at only 0.2 U/mL also a range of Epo concentrations (0, 0.2, 2, and 10 U/mL; Epo (Fig. 3C), although activation levels are decreased and P see Fig. 1F). There were no significant differences ( >0.05) slightly delayed, particularly for the STAT5 pathway. in invasion between 0 and 0.2 U/mL Epo in any of the three Rama cell lines. At 2 U/mL Epo, the invasion capacity of the Identification of Signaling Pathway–Mediating Rama 37-28 cells was significantly higher than both Rama Malignant Cell Behavior 37-50 and Rama 37 parental cells, whereas at 10 U/mL Kinase inhibitors were used to assess the roles of the ac- Epo, both the Rama 37-50 and Rama 37-28 had signifi- tivated signaling pathways in mediating Rama 37-28 cell cantly higher invasion capacity than Rama 37 parental cells. invasion. Concentrations of three inhibitors, JAK inhibitor Thus, invasion potential seems to be dependent on the 1(5μmol/L), PI-103 (0.5 μmol/L), and U0126 (20 μmol/L), combination of EpoR copy number and the concentration which inhibit JAK, PI3K (p110α subunit), and mitogen- of Epo used (Fig. 1F). EpoR mRNA levels in our models are activated protein kinase kinase, in the JAK2/STAT5, higher than in a range of cancer cell lines but comparable PI3K/Akt, and Ras/ERK pathways, respectively, were opti- with levels in human erythroid progenitors and a human mized to determine the lowest concentration required to erythroleukemia cell line, UT7, which have been reported abolish Epo-induced signaling as determined by immuno- ∼ to express 10,000 receptors per cell (Table 2; ref. 34). blotting but without significantly affecting the other pathways (Fig. 4A). The traditional JAK-STAT inhibitor Epo Receptor Knockdown AG-490 failed to abrogate levels of phosphoSTAT5 even Epo-induced increases in adhesion, migration, invasion, at a concentration of 100 μmol/L (data not shown). These and colony-forming potential in the Rama 37-28 EpoR- inhibitors were then included in the assessment of Rama transfected cell line suggest that EpoR overexpression alone 37-28 cell invasion. Inhibition of the PI3K/Akt and Ras/ can increase malignant potential. In this model, silencing ERK pathways abrogates invasion potential even in the of EpoR expression using RNAi indicated that changes in presence of Epo (Fig. 4B). Although the JAK2/STAT5 sig- cell behavior were mediated directly through the Epo stim- naling is effectively inhibited with a decrease in levels of ulation of EpoR. Specific siRNA-mediated knockdown of phosphorylated STAT5, the decrease in invasion in re- the EpoR in Rama 37-28 cells (Table 3; Fig. 2A) resulted sponse to Epo was not significant compared with Rama in a significant reduction in the invasive capacity of the 37-28 cells without inhibitor. In addition, the presence cells (Fig. 2B). To confirm the role of EpoR in malignant of Epo still conferred a significant increase in invasion cell behavior, the effects of Epo stimulation on a human compared with nontreated cells even in the presence of cell line expressing endogenous EpoR were analyzed. EpoR JAK inhibitor 1. However, inhibition of all three pathways mRNA and protein levels were assessed in parental MDA- led to a significant decrease in adhesion potential (Fig. 4C). MB-435s cells and, following siRNA-mediated EpoR knockdown, showed a significant reduction in receptor le- Discussion vels (Table 3; Fig. 2A). Parental MDA-MB-435s cells show a significant increase in invasion and adhesion capacity in The anemia of cancer is associated with poor prognosis, response to Epo stimulation (P < 0.05), which is abrogated shortened survival, and a reduced quality of life (35). Cor- following EpoR knockdown (P < 0.05; Fig. 2C). The re- rection of anemia by ESAs provides an attractive alternative sults suggest a role for Epo/EpoR in increasing malignant to regular blood transfusions but concerns about the use of cell behavior in a human cancer cell line. ESAs have been reported in clinical trials involving a range

618 Mol Cancer Res; 8(4) April 2010 Molecular Cancer Research

Erythropoietin-Induced Malignant Cell Behavior

FIGURE 1. EpoR overexpression. A, immunoblot analysis of EpoR protein expression. Cell lysates were diluted and 15 μg of protein were loaded on a SDS 8% (w/w) polyacrylamide gel. Specific proteins were detected using antibodies to EpoR or β-actin. B, invasion, (C) adhesion, (D) migration, and (E) colony formation capacity were measured in the presence/absence of Epo. Percentage invasion, migration, adhesion, and colony formation were calculated relative to Rama 37 (−Epo) control. Rama 37-28 cell malignant cell behavior was significantly increased in the presence of Epo compared with the Rama 37 parental cells using Matrigel-coated multiwell plates. Columns, mean from three independent experiments; bars, SD. F, invasion potential was also measured in three cell lines with varying levels of EpoR (low, Rama 37; medium, Rama 37-50; and high, Rama 37-28) at four different concentrations of Epo (0, 0.2, 2, and 10 U/mL).

www.aacrjournals.org Mol Cancer Res; 8(4) April 2010 619

Shi et al.

Table 2. Q-PCR analysis of EpoR mRNA expression in a range of cancer cell lines, with conversion to copy number per 10 ng RNA

Cells Cell type CT value EpoR copy number % compared with erythroid progenitors

Rama 37 Rat mammary epithelial >35 N/D N/D Rama 37-50 Rat mammary epithelial 22.8 ± 0.14 8,595 ± 1,303 28 Rama 37-28 Rat mammary epithelial 19.28 ± 0.43 133,652 ± 59,987 435 BFU-E Human erythroid progenitors 21.16 ± 0.34 30,753 ± 8,408 100 MDA-MB-435S Human mammary epithelial 25.03 ± 0.35 1,608 ± 429 5 MDA-MB-231 Human mammary epithelial 25.39 ± 0.23 1,205 ± 205 4 MCF7 Human mammary epithelial 25.68 ± 0.11 951 ± 75 3 MCF10A Human mammary epithelial 27.17 ± 0.32 310 ± 77 1 MCF10AT Human mammary epithelial 26.02 ± 0.30 745 ± 157 2 MCF10A-CA1a Human mammary epithelial 28.64 ± 0.59 104 ± 46 0.3 C9 Human cervical epithelial >35 N/D N/D H838 Human non–small cell lung carcinoma 26.30 ± 0.28 603 ± 134 2 H157 Human non–small cell lung carcinoma 25.31 ± 0.33 1294 ± 309 4 H23 Human adenocarcinoma 25.82 ± 0.29 870 ± 180 3 UT7 Human erythroleukemia (AML-M6) 21.91 ± 0.83 19,893 ± 10,642 65 K562 Human chronic myeloid leukemia 25.51 ± 0.69 1,210 ± 519 4

Abbreviation: N/D, nondetectable using 10 ng RNA template.

of cancers (reviewed in ref. 36). An inverse relationship be- adhesion, invasion, migration, and colony formation in tween EpoR expression and disease-free and overall surviv- Rama 37-28 cells, suggesting that the presence of functional al in breast cancer has been reported (37) and Epo may EpoR is capable of mediating transformation to a malignant only have a negative effect in cancers that are positive for cell phenotype. EpoR expression using immunohistochemistry (35). How- When invasion capacity was compared among Rama 37 ever, further studies are needed with more specific antibo- parental cells, Rama 37-28, and Rama 37-50 (a clone with dies and greater sample numbers to determine the an intermediate level of EpoR overexpression), no signifi- prognostic significance of EpoR expression on tumors. cant differences were found at 0 and at 0.2 U/mL Epo in The biological effects of Epo stimulation of tumor cells any of the three Rama cell lines. At 2 U/mL Epo, Rama 37 is still debated with reports of enhanced survival (8, 11), parental and Rama 37-50 were similar, but the invasion proliferation (4, 6, 38-40), resistance to treatment (38, capacity of Rama 37-28 was significantly higher than 41, 42), tumor angiogenesis (43-45), chemotaxis (46), in- Rama 37 parental cells. At 10 U/mL Epo, both Rama vasion (47-50), and migration (40, 46, 51). Others have 37-50 and Rama 37-28 have significantly higher invasion reported no discernible effects of Epo (52). capacity than Rama 37 parental cells. In this context, it is In the present study, we have examined the role of EpoR in mediating malignant cell behavior in a rat mammary epithelial cell line model. The parental Rama 37 cells with Table 3. Q-PCR analysis of EpoR mRNA ex- low endogenous EpoR expression display a benign phenotype. Generation of a modified cell line with stable expres- pression in Rama 37-28 cells, parental MDA- sion of EpoR (Rama 37-28) resulted in significant changes MB-435s cells, and following siRNA-mediated in cell signaling in response to Epo stimulation and the cre- knockdown ation of a cell line with a malignant phenotype. The increased EpoR expression in the Rama 37-28 cell Cell line CT value Knockdown efficiency line resulted in the activation of all three major signaling pathways downstream of EpoR in response to Epo with sig- Rama 37-28 (0 h) 19.9 nificant increases in levels of phospho-STAT5, phospho- Rama 37-28 + 22.0 72.4 ± 8.6% Akt, and phospho-ERK1/2 >60 min after Epo stimulation. siRNA (48 h) Parental Rama 37 cells showed only minor activation of MDA-MB-435s (0 h) 25.3 two signaling pathways and no significant change in cell MDA-MB-435s + 27.9 84.1% behavior in response to Epo, confirming their benign phe- siRNA (48 h) notype. In contrast, Epo treatment caused an increase in

620 Mol Cancer Res; 8(4) April 2010 Molecular Cancer Research

Erythropoietin-Induced Malignant Cell Behavior

FIGURE 2. RNAi-mediated EpoR knockdown. A, immunoblot analysis of EpoR mRNA expression in Rama 37-28 cells, parental MDA-MB-435s cells, and following siRNA-mediated knockdown. B, abrogated invasion capacity in response to Epo compared with the Rama 37-28 control cells. C, invasion and (D) adhesion capacity following Epo stimulation of endogenous EpoR in MDA-MB-435s and following EpoR knockdown. Columns, mean from three independent experiments; bars, SD. of interest to note that with patients receiving recombinant these intersecting biochemical networks may drive tumor- human Epo therapy, serum levels of ∼4 U/mL Epo have igenesis irrespective of normal biological cues. A range of been reported in patients receiving 120 U/kg Epo (53). signaling intermediate inhibitors were used to determine In our study, preliminary investigations with the human in- which pathways are involved in mediating Epo-induced vasive cell line MDA-MB-435s, which expresses endogenous malignant cell behavior. EpoR, shows a significant increase in invasion and adhesion in PI3Ks are lipid kinases that are activated by receptor response to Epo. This is clearly mediated through EpoR as tyrosine kinases and other cell surface receptors to synthesize siRNA-mediated receptor knockdown resulted in significant the lipid second messenger phosphatidylinositol-3,4,5- attenuation of the invasion and adhesion potential in the trisphosphate. Phosphatidylinositol-3,4,5-trisphosphate acts presence of Epo compared with parental controls. These re- as a docking site at the plasma membrane to recruit and ac- sults confirm in a human model system that Epo stimulation tivate proteins containing phospholipid-binding domains of endogenous EpoR promotes malignant cell behavior. such as Akt, a serine-threonine kinase that signals to suppress Multiple signaling pathways are activated by Epo bind- apoptosis and promote cell growth. The PI3K pathway is fre- ing to the EpoR including the JAK/STAT, PI3K/Akt, and quently activated in cancer due to amplification or gain-of- Ras/ERK cascades in our mammary epithelial model. Ac- function mutations of the PIK3CA gene, which encodes tivation of all three signaling cascades through other class I the p110α subunit of Class 1 PIK, loss of function of PTEN, cytokine receptors (e.g., prolactin and leptin) in breast can- which catalyses the conversion of phosphatidylinositol-3,4,5- cer has been previously reported (54, 55). Activation of trisphosphate to phosphatidylinositol 4,5-bisphosphate, or

www.aacrjournals.org Mol Cancer Res; 8(4) April 2010 621

Shi et al.

FIGURE 3. Signaling pathway activation in response to Epo stimulation. Epo-induced signaling was analyzed (A) in Rama 37 parental cells at 10 U/mL Epo, (B) in Rama 37-28 cells at 10 U/mL Epo, and (C) in Rama 37-28 cells at 0.2 U/mL Epo. Protein lysates were analyzed on a 10% SDS polyacrylamide gel. Specific proteins were detected using antibodies to phospho-Stat5, Stat5, phospho-Akt, Akt, phospho-ERK1/2, ERK1/2, and β-actin. Bands were quantified using densitometric analysis and normalized to β-actin. A representative blot from triplicate experiments is shown. Solid box, maximum signaling at 15 min. Dashed box, time-matched−Epo controls.

622 Mol Cancer Res; 8(4) April 2010 Molecular Cancer Research

Erythropoietin-Induced Malignant Cell Behavior

FIGURE 4. The effect of signaling pathway inhibition on cell behavior. A, inhibition of JAK2/STAT5, PI3K/Akt, and Ras/ERK pathways using 5 μmol/L JAK inhibitor 1, 0.5 μmol/L PI-103, and 20 μmol/L U0126 small-molecule inhibitors, respectively. The cells were pretreated with inhibitor for 1 h before the addition of Epo. At these concentrations, only the targeted pathway showed significantly reduced signaling capacity. B and C, cell behavior analyzed in the presence of the small-molecule inhibitors. Percentage (B) invasion and (C) adhesion were calculated relative to Rama 37-28 (−Epo) controls. SDs from triplicate experiments are shown and t tests were used to determine statistically significant differences.

www.aacrjournals.org Mol Cancer Res; 8(4) April 2010 623

Shi et al.

aberrant growth factor or integrin receptor signaling (re- of a JAK inhibitor, was still capable of eliciting a significant viewed in ref. 56). Thus, the PI3K pathway has emerged as increase in invasion when compared with controls. How- an attractive target for small-molecule inhibitors with thera- ever, inhibition of JAK/STAT signaling resulted in abrogat- peutic potential for disrupting the initiation and progression ing the adhesion potential in Rama 37-28 cells, suggesting of cancers. The arylmorpholine agent PI-103 inhibits both that this signaling cascade may play different roles in pro- PI3K and mammalian target of rapamycin, enabling the in- moting malignant cell behavior. hibition of both the forward signal from PI3K and the nega- Growth factors stimulate cells to take up excess nutri- tive feedback from mammalian target of rapamycin. PI-103 ents, and to use them for anabolic processes and activation inhibits invasion in breast and ovarian cancer xenograft mod- of their signaling pathways can play a major role in medi- els (57) and induces proliferative arrest of glioma cells, mediating changes in cell metabolism (62). An altered ated through its dual inhibitory action (58). In the Rama 37- metabolic phenotype with a switch to aerobic glycolysis 28 cell model, a low concentration of PI-103 (0.5 μmol/L) is one of the hallmarks of tumorigenesis first noted by Otto effectively abolished the phosphorylation of the PI3K down- Warburg (63). Akt, one of the most frequently activated stream target Akt and resulted in a significant abrogation of protein kinases in human malignancy, which transduces invasion and adhesion potential, even in the presence of Epo. growth factor effects on cell survival, growth, and prolifer- The Ras/ERK pathway couples signals from cell surface ation, is also a critical mediator of accelerated glycolytic receptors to a wide range of cellular processes including pro- and oxidative metabolism (64). Our present data indicate liferation, differentiation, apoptosis, and cell cycle progres- that stimulation of EpoR by pharmacologic levels of Epo sion (59). Amplification of ras genes and activating leads to rapid and sustained activation of Akt, underscoring mutations in this pathway occur frequently in cancer, lead- the key role played by this pathway in the transformation ing to constitutive signaling activation (reviewed in ref. 60). from a benign to a tumorigenic cell phenotype. RasisasmallGTP-bindingproteinthatliesupstreamof Breast cancer progression depends not only on primary both the Raf/mitogen-activated protein kinase kinase/ERK tumor growth but also on the ability of tumor cells to me- and PI3K pathways. It is activated by growth factor/mitogen tastasize to distant sites; thus, the invasion, adhesion, and binding to their reciprocal receptor, which leads to the asso- migration capacities increased by Epo activation of the ciation of the Src homology and collagen/growth factor EpoR in our study represents an additional mechanism receptor binding protein 2/SOS complex and the subse- for induction of pathways leading to a metastatic pheno- quent conformational activation of Ras. Activation of the type. Clearly, this is only one model system and further Ras-ERK and mitogen-activated protein kinase/c-Jun- experiments in vivo are required to fully understand the NH2-kinase pathways has been implicated in mediating role of the Epo receptor in malignant progression. Whether proliferation and migration of breast cancer cells (40). In Epo is capable of accelerating tumor growth in cancer our study, exposure of EpoR expressing Rama 37-28 cells patients remains an open question. to Epo in the presence of the specific mitogen-activated protein kinase kinase inhibitor U0126 resulted in the complete abolition of the Epo-induced induction of ERK1/2 phos- Disclosure of Potential Conflicts of Interest phorylation. This in turn eliminated the invasion and adhe- No potential conflicts of interest were disclosed. sion capacity of this cell line despite Epo stimulation. JAK/STAT inhibitors are currently used to treat hematological malignancies (reviewed in ref. 61). However, the Acknowledgments role of inhibitors of this pathway in solid tumors is less well The authors would like to thank Susan Price for excellent technical assistance understood. Epo-induced invasion of head and neck squa- with growing erythroid progenitors. mous cell carcinoma through the JAK/STAT pathway has The costs of publication of this article were defrayed in part by the payment of been reported (46) but our results suggest that this path- page charges. This article must therefore be hereby marked advertisement in way does not play a major role in mediating invasion of accordance with 18 U.S.C. Section 1734 solely to indicate this fact. mammary epithelial cells. Although a decrease in invasion Received 07/06/2009; revised 02/02/2010; accepted 03/01/2010; published capacity was shown, Epo stimulation, even in the presence OnlineFirst 03/30/2010.

References 1. Hodges VM, Rainey S, Lappin TR, Maxwell AP. Pathophysiology of 6. Acs G, Acs P, Beckwith SM, et al. Erythropoietin and erythropoietin anemia and erythrocytosis. Crit Rev Oncol Hematol 2007;64:139–58. receptor expression in human cancer. Cancer Res 2001;61:3561–5. 2. Jelkmann W. Erythropoietin after a century of research: younger than 7. Acs G, Zhang PJ, Rebbeck TR, Acs P, Verma A. Immunohistochem- ever. Eur J Haematol 2007;78:183–205. ical expression of erythropoietin and erythropoietin receptor in breast 3. Lappin TR, Maxwell AP, Johnston PG. EPO's alter ego: erythropoie- carcinoma. Cancer 2002;95:969–81. tin has multiple actions. Stem Cells 2002;20:485–92. 8. Acs G, Zhang PJ, McGrath CM, et al. Hypoxia-inducible erythropoi- 4. Westenfelder C, Baranowski RL. Erythropoietin stimulates prolifera- etin signaling in squamous dysplasia and squamous cell carcinoma tion of human renal carcinoma cells. Kidney Int 2000;58:647–57. of the uterine cervix and its potential role in cervical carcinogenesis 5. Selzer E, Wacheck V, Kodym R, et al. Erythropoietin receptor expres- and tumor progression. Am J Pathol 2003;162:1789–806. sion in human melanoma cells. Melanoma Res 2000;10:421–26. 9. Eccles TG, Patel A, Verma A, et al. Erythropoietin and the erythropoietin

624 Mol Cancer Res; 8(4) April 2010 Molecular Cancer Research

Erythropoietin-Induced Malignant Cell Behavior

receptor are expressed by papillary thyroid carcinoma from children 33. Kurisetty VV, Johnston PG, Johnston N, et al. RAN GTPase is an ef- and adolescents. Expression of erythropoietin receptor might be a fector of the invasive/metastatic phenotype induced by osteopontin. favorable prognostic indicator. Ann Clin Lab Sci 2003;33:411–22. Oncogene 2008;27:7139–49. 10. Yasuda Y, Fujita Y, Matsuo T, et al. Erythropoietin regulates tumour 34. Komatsu N, Yamamoto M, Fujita H, et al. Establishment and charac- growth of human malignancies. Carcinogenesis 2003;24:1021–9. terization of an erythropoietin-dependent subline, UT-7/Epo, derived 11. Acs G, Xu X, Chu C, Acs P, Verma A. Prognostic significance of from human leukemia cell line, UT-7. Blood 1993;82:456–64. erythropoietin expression in human endometrial carcinoma. Cancer 35. Henke M, Mattern D, Pepe M, et al. Do erythropoietin receptors on 2004;100:2376–86. cancer cells explain unexpected clinical findings? J Clin Oncol 2006; 12. Arcasoy MO, Amin K, Vollmer RT, Jiang X, Demark-Wahnefried W, 24:4708–13. Haroon ZA. Erythropoietin and erythropoietin receptor expression 36. Newland AM, Black CD. Tumor progression associated with erythro- in human prostate cancer. Mod Pathol 2005;18:421–30. poiesis-stimulating agents. Ann Pharmacother 2008;42:1865–70. 13. Dagnon K, Pacary E, Commo F, et al. Expression of erythropoietin 37. Pelekanou V, Kampa M, Kafousi M, et al. Erythropoietin and its re- and erythropoietin receptor in non-small cell lung carcinomas. Clin ceptor in breast cancer: correlation with steroid receptors and out- Cancer Res 2005;11:993–9. come. Cancer Epidemiol Biomarkers Prev 2007;16:2016–23. 14. Shenouda G, Mehio A, Souhami L, et al. Erythropoietin receptor ex- 38. Pajonk F, Weil A, Sommer A, Suwinski R, Henke M. The erythropoi- pression in biopsy specimens from patients with uterine cervix squa- etin-receptor pathway modulates survival of cancer cells. Oncogene mous cell carcinoma. Int J Gynecol Cancer 2006;16:752–6. 2004;23:8987–91. 15. Elliott S, Busse L, Bass MB, et al. Anti-Epo receptor antibodies do 39. Feldman L, Wang Y, Rhim JS, Bhattacharya N, Loda M, Sytkowski not predict Epo receptor expression. Blood 2006;107:1892–5. AJ. Erythropoietin stimulates growth and STAT5 phosphorylation in 16. Brown WM, Maxwell P, Graham AN, et al. Erythropoietin Receptor human prostate epithelial and prostate cancer cells. Prostate 2006; Expression in non-small cell lung carcinoma: a question of antibody 66:135–45. specificity. Stem Cells 2007;25:718–22. 40. Fu P, Jiang X, Arcasoy MO. Constitutively active erythropoietin re- 17. Henke M, Laszig R, Rübe C, et al. Erythropoietin to treat head and neck ceptor expression in breast cancer cells promotes cellular prolifera- cancer patients with anaemia undergoing radiotherapy: randomised, tion and migration through a MAP-kinase dependent pathway. double-blind, placebo-controlled trial. Lancet 2007;362:1225–60. Biochem Biophys Res Commun 2009;379:696–701. 18. Leyland-Jones B, BEST Investigators and Study Group. Breast can- 41. Belenkov AI, Shenouda G, Rizhevskaya E, et al. Erythropoietin in- cer trial with erythropoietin terminated unexpectedly. Lancet Oncol duces cancer cell resistance to ionizing radiation and to cisplatin. 2003;4:459–60. Mol Cancer Ther 2004;3:1525–32. 19. Bohlius J, Wilson J, Seidenfeld J, et al. Erythropoietin or darbepoetin 42. Temkin SM, Hellmann M, Serur E, Lee YC, Abulafia O. Erythropoietin for patients with cancer. Cochrane Database Syst Rev 2006;3: administration during primary treatment for locally advanced cervical CD003407. carcinoma is associated with poor response to radiation. Int J Gyne- 20. Bohlius J, Wilson J, Seidenfeld J, et al. Recombinant human erythro- col Cancer 2006;16:1855–61. poietins and cancer patients: updated meta-analysis of 57 studies 43. Ribatti D, Presta M, Vacca A, et al. Human erythropoietin induces a including 9353 patients. J Natl Cancer Inst 2006;98:708–14. pro-angiogenic phenotype in cultured endothelial cells and stimu- 21. Lappin TR, Maxwell AP, Johnston PG. Warning Flags for erythropoi- lates neovascularization in vivo. Blood 1999;93:2627–36. esis stimulating agents and cancer associated anemia. Oncologist 44. Hardee ME, Cao Y, Fu P, et al. Erythropoietin blockade inhibits the 2007;12:362–5. induction of tumor angiogenesis and progression. PLoS ONE 2007; 22. U S. Food and Drug Administration. Information for Healthcare Profes- 2:e549. sionals. Erythropoiesis Stimulating Agents (ESA) (Aranesp (darbepoetin), 45. Okazaki T, Ebihara S, Asada M, Yamanda S, Niu K, Arai H. Erythro- Epogen (epoetin alfa), and Procrit (epoetin alfa)). FDA Alert, poietin promotes the growth of tumors lacking its receptor and de- February 16, 2007. Available at: http://www.fda.gov/Drugs/DrugSafety/ creases survival of tumor-bearing mice by enhancing angiogenesis. PostmarketDrugSafetyInformationforPatientsandProviders/ Neoplasia 2008;10:932–9. ucm126485.htm. 46. Hamadmad SN, Hohl RJ. Erythropoietin stimulates cancer cell mi- 23. Rizzo JD, Seidenfeld J, Piper M, Aronson N, Lichtin A, Littlewood TJ. gration and activates RhoA protein through a mitogen-activated pro- Erythropoietin: a paradigm for the development of practice guide- tein kinase/extracellular signal-regulated kinase-dependent lines. Hematology (Am Soc Hematol Educ Program) 2001:10–30. mechanism. J Pharmacol Exp Ther 2008;324:1227–33. 24. Bohlius J, Schmidlin K, Brillant C, et al. Recombinant human eryth- 47. Lai SY, Childs EE, Xi S, et al. Erythropoietin-mediated activation of ropoiesis-stimulating agents and mortality in patients with cancer: a JAK-STAT signaling contributes to cellular invasion in head and neck meta-analysis of randomised trials. Lancet 2009;373:1532–42. squamous cell carcinoma. Oncogene 2005;24:4442–9. 25. Tonelli M, Hemmelgarn B, Reiman T, et al. Benefits and harms of 48. Mohyeldin A, Lu H, Dalgard C, et al. Erythropoietin signaling pro- erythropoiesis-stimulating agents for anemia related to cancer: a motes invasiveness of human head and neck squamous cell carci- meta-analysis. Can Med Assoc J 2009;180:E62–71. noma. Neoplasia 2005;7:537–43. 26. Dunlop EA, Percy MJ, Boland MP, Maxwell AP, Lappin TR. Induction 49. Mohyeldin A, Dalgard CL, Lu H, et al. Survival and invasiveness of of signaling in non-erythroid cells by pharmacological levels of eryth- astrocytomas promoted by erythropoietin. J Neurosurg 2007;106: ropoietin. Neurodegen Dis 2006;3:94–100. 338–50. 27. Dunlop EA, Maxwell AP, Lappin TR. Impaired downregulation follow- 50. Paragh G, Kumar SM, Rakosy Z, Choi SC, Xu X, Acs G. RNA ing erythropoietin receptor activation in non-small cell lung carcino- interference-mediated inhibition of erythropoietin receptor expres- ma. Stem Cells 2007;25:380–4. sion suppresses tumor growth and invasiveness in A2780 human 28. Dunnington DJ, Hughes CM, Monaghan P, Rudland PS. Phenotypic ovarian carcinoma cells. Am J Pathol 2009;74:1504–14. instability of rat mammary tumor epithelial cells. J Natl Cancer Inst 51. Lester RD, Jo M, Campana WM, Gonias SL. Erythropoietin promotes 1983;71:1227–40. MCF-7 breast cancer cell migration by an ERK/mitogen-activated 29. Cailleau R, Olivé M, Cruciger QV. Long-term human breast carcino- protein kinase-dependent pathway and is primarily responsible for ma cell lines of metastatic origin: preliminary characterization. In Vitro the increase in migration observed in hypoxia. J Biol Chem 2005; 1978;14:911–5. 280:39273–7. 30. Chambers AF. MDA-MB-435 and M14 cell lines: identical but not 52. LaMontagne KR, Butler J, Marshall DJ, et al. Recombinant epoetins M14 melanoma? Cancer Res 2009;69:5292–3. do not stimulate tumor growth in erythropoietin receptor-positive 31. El-Tanani MK, Campbell FC, Crowe P, et al. BRCA1 suppresses osteo- breast carcinoma models. Mol Cancer Ther 2006;5:347–55. pontin-mediated breast cancer. J Biol Chem 2006;281:26587–601. 53. Macdougall IC, Neubert P, Coles GA, et al. Pharmacokinetics 32. Thompson A, Quinn MF, Grimwade D, et al. Global down-regulation of of recombinant human erythropoietin in patients on continuous HOX gene expression in PML-RARα+ acute promyelocytic leukemia ambulatory peritoneal dialysis. Lancet 1989;333:425–7. identified by small-array real-time PCR. Blood 2003;101:1558–65. 54. Neilson LM, Zhu J, Xie J, et al. Coactivation of janus tyrosine kinase

www.aacrjournals.org Mol Cancer Res; 8(4) April 2010 625

Shi et al.

(Jak)1 positively modulates prolactin-Jak2 signaling in breast cancer: 59. McCubrey JA, Steelman LS, Chappell WH, et al. Roles of the Raf/ recruitment of ERK and signal transducer and activator of transcrip- MEK/ERK pathway in cell growth, malignant transformation and drug tion (Stat)3 and enhancement of Akt and Stat5a/b pathways. Mol resistance. Biochim Biophys Acta 2007;1773:1263–84. Endocrinol 2007;21:2218–32. 60. Karnoub AE, Weinberg RA. Ras oncogenes: split personalities. 55. Cirillo D, Rachiglio AM, la Montagna R, Giordano A, Normanno N. Nat Rev Mol Cell Biol 2008;9:517–31. Leptin signaling in breast cancer: an overview. J Cell Biochem 61. Ferrajoli A, Faderl S, Ravandi F, Estrov Z. The JAK-STAT pathway: a 2008;105:956–64. therapeutic target in hematological malignancies. Curr Cancer Drug 56. Dillon RL, White DE, Muller WJ. The phosphatidyl inositol 3-kinase Targets 2006;6:671–9. signaling network: implications for human breast cancer. Oncogene 62. Christofk HR, Vander Heiden MG, Wu N, Asara JM, Cantley LC. 2007;26:1338–45. Pyruvate kinase M2 is a phosphotyrosine-binding protein. Nature 57. Raynaud FI, Eccles S, Clarke PA, et al. Pharmacologic characteriza- 2008;452:181–6. tion of a potent inhibitor of class I phosphatidylinositide 3-kinases. 63. Warburg O. On the origin of cancer cells. Science 1956;123: Cancer Res 2007;67:5840–50. 309–14. 58. Fan QW, Knight ZA, Goldenberg DD, et al. A dual PI3 kinase/mTOR 64. Robey RB, Hay N. Is Akt the “Warburg kinase”?—Akt-energy inhibitor reveals emergent efficacy in glioma. Cancer Cell 2006;9: metabolism interactions and oncogenesis. Semin Cancer Biol 2009; 341–9. 19:25–31.

626 Mol Cancer Res; 8(4) April 2010 Molecular Cancer Research

Erythropoietin-Induced Activation of the JAK2/STAT5, PI3K/Akt, and Ras/ERK Pathways Promotes Malignant Cell Behavior in a Modified Breast Cancer Cell Line

Zhanzhong Shi, Vivien M. Hodges, Elaine A. Dunlop, et al.

Mol Cancer Res 2010;8:615-626. Published OnlineFirst March 30, 2010.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-09-0264

Cited articles This article cites 62 articles, 19 of which you can access for free at: http://mcr.aacrjournals.org/content/8/4/615.full#ref-list-1

Citing articles This article has been cited by 4 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/8/4/615.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mcr.aacrjournals.org/content/8/4/615. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.